基于矩阵逼近的模型修正方法的研究

提出一种新的以试验振动和参数辨识的数据为参数，进行有限元分析模型修正的方法。该方法基于矩阵最佳逼近理论，运用Ｂａｙｅｓ估计原理来处理试验结果误差带来的试验模态可信度问题，求取分析模对试验获得的不完备模态的谱点的最佳逼近结果，最后获得质量阵的最小修正模型。

RESEARCH OF A MODEL UPDATING METHOD BASED ON MATRIX APPROXIMATION THEORY

A new method for updating FE analytical model is put forward in this paper. In this method, the result data from experimental test and modal parametric identification are used as reference base to refine the analytical model. The optimal approximation is defined as the refinement of FE model with minimum mass matrix change to make the spectrum paint of updated model approach to the one coming from the incomplete experimental mode. The optimum matrix approach theory introduced in this method is testified to have the abthty keeping the misshapen mode away, which often come up in other incomplete mode updating methods. As another key barrier in practical updating process, the experiment errors are taken into aecount by making use of Baysian estimation principle. An example of cantilever shows the physical characteristics of this methed and the clearness of computation. This method needs no iteration.